WITHDRAWN: Deviation Error: assessing machine learning predictions for replicate measurements in genomics and beyond

Imagine you are a detective trying to solve a mystery, but instead of looking for fingerprints, you are looking for patterns in the tiny instruction manuals inside every living cell. This field is called genomics, and it's like trying to read a library of billions of books to understand how life works. But here's the tricky part: when scientists try to read these books, they often make copies of the same page to be sure they got it right. These copies are called "replicate measurements." The big question in this world is: how do we know if our guesses about what's in those books are actually good? We need a way to measure the "mistake" or the "deviation" between what we predicted and what we actually saw. If our measuring tape is broken, we might think we've found a new species of dragon when we've just misread a lizard. That's why having a perfect ruler to measure these mistakes is the most important tool a scientist can have.

Now, picture a team of scientists named Husam and J. Timothy who decided to invent a brand-new, super-precise ruler for this job. They called their invention the "Deviation Error." They wanted to use this new ruler to check how well scientists could predict the results of those duplicate measurements in genomics and other fields. It sounded like a brilliant idea, a shiny new tool that would help everyone get their math right.

However, the story takes a sudden turn. After looking at their own invention very closely, Husam and J. Timothy realized that their new ruler wasn't quite ready for the toolbox yet. They discovered that the math behind the "Deviation Error" needed a massive amount of extra work to be proven as a "proper scoring rule"—which is just a fancy way of saying the math needs to be rock-solid and unbreakable before it can be trusted to measure anything important. They found that their current version was a bit wobbly, like a ladder that hasn't been bolted to the wall yet.

Because of this, the authors made a very responsible decision: they have officially withdrawn this paper. They are essentially saying, "We built this cool idea, but we need to go back to the workshop and do some serious heavy lifting to fix the math and rebuild our test cases." They plan to come back to this project in the future once they have done the necessary research to make the Deviation Error truly reliable. For now, they are explicitly asking everyone to not cite this work as a reference because it has been withdrawn and is not finished. So, while the idea of a better measuring tool is still on the table, this particular attempt is being paused until the math can stand up on its own two feet.